A funding change moved 98 NZ children off their insulin pumps. No control arm

In October 2024, New Zealand's pharmaceutical funding agency PHARMAC introduced fully funded access to Tandem t:slim X2 and Ypsomed YpsoPump insulin pumps, alongside Abbott and Dexcom continuous glucose monitoring [s1]. Medtronic continuous glucose monitoring remained unfunded, which meant continuing to use a Medtronic automated insulin delivery system now carried substantial personal cost — and prompted a prioritised transition of all paediatric Medtronic users at one centre onto funded alternatives [s1].

A retrospective audit of what happened next has been published in Diabetes Technology & Therapeutics. It is a small, uncontrolled, single-centre study, and the most useful thing about it is what that combination can and cannot demonstrate about a national funding decision.

What the audit found

Ninety-eight children and adolescents transitioned automated insulin delivery systems, with complete six-month outcome data available for 79 — mean age 11.6 ± 3.0 years, diabetes duration 4.68 years, pump use duration 2.44 years [s1]. Most moved to Tandem Control-IQ (n=66) or YpsoPump with CamAPS FX (n=11) [s1].

Mean HbA1c decreased by 0.24%, from 7.14 ± 0.79% to 6.87 ± 0.86% (P=0.0012), with no episodes of diabetic ketoacidosis or level 3 hypoglycaemia recorded [s1]. The authors conclude that funding-driven transitions between automated insulin delivery systems can be safely implemented with structured education and follow-up [s1].

One small inconsistency is worth flagging for readers who check arithmetic: the reported mean decrease of 0.24% does not exactly match the difference between the two reported group means, which the summary does not explain — a discrepancy of the kind produced by paired analysis on a subset, or by rounding.

Why the design limits the claim

The audit has no control group. Every one of the 79 children switched, so there is no comparable set who stayed on their original system over the same six months [s1]. Anything that changes HbA1c over a half-year in a paediatric diabetes clinic — seasonal variation, a clinic-wide education push, regression to the mean among those who started higher — is indistinguishable from a device effect here.

The transition also came with structured education and follow-up [s1]. That is good clinical practice and it is also a co-intervention. A cohort that receives concentrated attention from a diabetes team tends to improve; the audit cannot separate the new pumps from the contact hours that accompanied them.

Complete six-month data were available for 79 of the 98 who transitioned [s1]. Loss on that scale is unremarkable for a real-world audit, but it runs in an unknown direction: children who struggled with the new system are plausibly among those who did not complete follow-up, and the result would look better than the full cohort's experience if so.

Finally, the reported effect is small. A 0.24% fall in mean HbA1c [s1] is at the low end of what is usually treated as clinically meaningful, and the cohort started at 7.14% — already close to target.

The number that puts the cohort in context

That starting point is what makes the national benchmarking audit published the previous October worth reading alongside it.

KiwiDiab collated all cases of type 1 and type 2 diabetes from the 13 regional centres providing care in New Zealand for the 2023 calendar year, identifying 1,849 children and adolescents with diabetes, of whom 1,642 (88.8%) had type 1 diabetes [s2]. Overall, 23% of those with type 1 diabetes achieved an HbA1c below 53 mmol/mol (7.0%) [s2].

Mean HbA1c differed sharply by treatment modality: 59 ± 11 mmol/mol (7.5% ± 1.0%) for any insulin pump use, 72 ± 19 (8.7% ± 1.8%) for multiple daily injections, and 96 ± 25 (11.0% ± 2.3%) for twice-daily insulin (p<0.001) [s2].

The transitioning cohort's baseline of 7.14% [s1] therefore sits below the national mean for children on any pump [s2]. This was not a struggling group being rescued by better technology; it was a group already doing comparatively well, which is the population in which a device change is least likely to show a large benefit — and, correspondingly, the population in which "no DKA and no level 3 hypoglycaemia" [s1] is least surprising.

The equity finding the single-centre audit cannot see

The national audit records something the transition study is not designed to capture. Among the 23% of children using automated insulin delivery, mean HbA1c was similar across ethnic groups [s2]. But among those on multiple daily injections or twice-daily insulin, HbA1c was higher for Pacific and Māori children than for European or Asian children (p<0.001) [s2].

The cohort overall was 69% NZ European, 16% Māori and 8.3% Pacific [s2]. Type 2 diabetes, by contrast, accounted for 151 of 1,849 children (8.2%), of whom 38% were Māori and 42% Pacific [s2].

Read together, those figures describe technology access as the point at which an outcome gap opens or closes. A funding decision that changes which devices are free is therefore an equity instrument whether or not it is framed as one — and the audit of 79 children at one centre says nothing about whether the October 2024 change widened or narrowed that gap.

What this evidence supports

The defensible reading is narrow. At one New Zealand centre, a forced switch between automated insulin delivery systems was carried out in 98 children without recorded diabetic ketoacidosis or severe hypoglycaemia in the 79 followed for six months, and mean HbA1c did not deteriorate [s1]. That is a safety signal, not an effectiveness result, and it applies to a well-controlled cohort supported by structured education.

What would answer the larger question is a national analysis using the KiwiDiab benchmarking infrastructure that already exists [s2] — all 13 centres, before and after October 2024, stratified by ethnicity and treatment modality. The dataset for that appears to be in place. The audit published here is one centre's contribution to it.